← Clinical Trials
Head and Neck Cancer Phase 3 Time-to-Event Analysis NCT02952586

JAVELIN Head and Neck 100: Complete Statistical Analysis of Avelumab in Squamous Cell Carcinoma of the Head and Neck

An independent statistical review of the randomized phase 3 JAVELIN Head and Neck 100 trial comparing avelumab plus standard of care chemoradiotherapy with placebo plus standard of care chemoradiotherapy for definitive treatment in patients with squamous cell carcinoma of the head and neck.

Phase 3  ·  Randomized  ·  Parallel  ·  Quadruple-masked  ·  Enrollment 697
Scope of this analysis

This page separates reported trial results from statistical interpretation. Numerical results and trial-specific design facts are limited to the ClinicalTrials.gov data posted on ClinicalTrials.gov for NCT02952586. Where the registry does not provide a requested detail, it is not inferred from other sources.

Registry note: This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.

1. Trial at a Glance

JAVELIN Head and Neck 100 was a randomized, parallel-group, quadruple-masked phase 3 trial evaluating avelumab in combination with standard of care chemoradiotherapy versus placebo with standard of care chemoradiotherapy for definitive treatment of squamous cell carcinoma of the head and neck. The registry reports 697 participants enrolled and a primary endpoint based on progression-free survival.

697
Enrolled
2 treatment groups
3
Phase
Phase 3
1
Primary endpoint
Time-to-event
5
Analyses posted
1 primary, 4 secondary
FeatureJAVELIN Head and Neck 100
PhasePhase 3
ConditionSquamous Cell Carcinoma of the Head and Neck
AllocationRandomized
Design modelParallel
MaskingQuadruple
Primary purposeTreatment
Enrollment697
InterventionsAvelumab and chemoradiation
Results postedYes
Outcome measures posted25
Statistical analyses posted5
Trial statusTerminated

2. Clinical Question

The central statistical question was whether adding avelumab to standard of care chemoradiotherapy changed progression-free survival compared with standard of care chemoradiotherapy alone in patients with squamous cell carcinoma of the head and neck.

Population

Patients with squamous cell carcinoma of the head and neck enrolled in the phase 3 randomized trial.

Intervention

Avelumab plus standard of care chemoradiotherapy.

Comparator

Placebo plus standard of care chemoradiotherapy.

Primary question

Does avelumab plus standard of care chemoradiotherapy improve investigator-assessed progression-free survival compared with placebo plus standard of care chemoradiotherapy?

3. Trial Design

01
Randomize697 enrolled
02
Two groupsAvelumab or placebo with SOC CRT
03
FollowTime-to-event outcomes
04
AssessInvestigator-assessed endpoints
05
CompareLog-rank and CMH methods
ARM A

Avelumab + Standard of Care Chemotherapy

  • Avelumab
  • Standard of care chemoradiotherapy
  • Time-to-event efficacy outcomes assessed in the full analysis set
ARM B

Placebo + SOC CRT

  • Placebo
  • Standard of care chemoradiotherapy
  • Time-to-event efficacy outcomes assessed in the full analysis set

The ClinicalTrials.gov record describes the study as randomized, parallel, and quadruple-masked. These design features are important statistically because randomization creates the basis for comparing outcomes between treatment assignments, while masking can reduce the influence of treatment knowledge on trial conduct and assessment.

Trial timeline

2016-11-28

Study start

The registry lists November 28, 2016 as the trial start date.

2019-12-23

Primary completion

The registry lists December 23, 2019 as the primary completion date.

Current registry status

Terminated

The ClinicalTrials.gov record identifies the study status as terminated.

4. Endpoints

The registry identifies one registered primary endpoint and reports five formal statistical analyses in total: one primary analysis and four secondary analyses.

EndpointRegistry definition / time frameAnalysis
Primary: Progression-free Survival (PFS) Progression-free Survival (PFS) Per Modified Response Evaluation Criteria in Solid Tumors v1.1 (RECIST v1.1) as Assessed by Investigator. From randomization until documented PD or death, censored date, whichever occurred first (up to 37 months). Kaplan-Meier method; log-rank comparison; hazard ratio
Secondary: Overall Survival (OS) From randomization to the date of death or censored date, whichever occurred first (up to 37 months). Log-rank comparison; hazard ratio
Secondary: Time to Locoregional Failure Time to Locoregional Failure Per Modified RECIST v1.1 as Assessed by Investigator. From the date of randomization to the date of the first documentation of locoregional recurrence or death, whichever occurred first. Log-rank comparison; hazard ratio
Secondary: Objective Response Rate (ORR) Objective Response Rate (ORR) Per Modified RECIST v1.1 as Assessed by Investigator. From randomization until disease progression or death, whichever occurred first (up to 37 months). Cochran-Mantel-Haenszel test; odds ratio
Secondary: Time to Distant Metastatic Failure Time to Distant Metastatic Failure Per Modified RECIST v1.1 as Assessed by Investigator. From the date of randomization to the date of the first documentation of distant metastatic or death (up to 37 months). Log-rank comparison; hazard ratio
Registry wording: The ClinicalTrials.gov record preserve the registry's endpoint terminology and time frames. The Time to Locoregional Failure and Time to Distant Metastatic Failure definitions are reproduced only to the extent provided in the ClinicalTrials.gov record.

5. Statistical Methodology

Full analysis set

The registry defines the analysis population for all five reported statistical analyses as the full analysis set (FAS), which included all randomized participants. This is important because the efficacy comparison remains anchored to treatment assignment rather than being restricted to participants who completed treatment or were fully evaluable after randomization.

Kaplan-Meier estimation

The primary PFS definition states that analysis was performed using the Kaplan-Meier method. Kaplan-Meier estimation is designed for time-to-event data in which some participants may not have experienced the event by the end of their observed follow-up. Such observations are censored rather than treated as though the event occurred at the censoring time.

Conceptual Kaplan-Meier estimator
S(t) = ∏ti ≤ t (1 − di/ni)

Here, di represents the number of events at an event time and ni represents the number at risk immediately before that time.

Log-rank testing

The primary PFS analysis and three of the four secondary analyses used the log-rank test. This is a standard approach for comparing time-to-event distributions between randomized groups across follow-up rather than comparing a single time point.

Hazard ratios

The primary PFS, OS, time to locoregional failure, and time to distant metastatic failure analyses all reported hazard ratios. A hazard ratio summarizes the relative event rate between groups within the time-to-event analysis framework.

Conceptual interpretation
HR = treatment-group hazard / comparator-group hazard

An HR of 1 corresponds to equal estimated hazards. An HR above 1 indicates a higher estimated hazard in the numerator group, while an HR below 1 indicates a lower estimated hazard, subject to the model and analysis assumptions.

Cochran-Mantel-Haenszel testing

The objective response rate analysis used a Cochran-Mantel-Haenszel test and reported an odds ratio. Unlike PFS and OS, ORR is a binary endpoint: participants are classified according to whether they met the response definition during the specified assessment period.

6. Results: Primary Endpoint

The registry reports a formal statistical analysis for the primary endpoint, investigator-assessed PFS. The comparison used the full analysis set and a log-rank test, with the treatment effect summarized by a hazard ratio.

Progression-Free Survival

Hazard ratio for progression or death

1.21

95% CI: 0.928–1.573   ·   P = 0.9199

Two-sided confidence interval  ·  Superiority hypothesis

Primary endpointAvelumab + SOC CRTPlacebo + SOC CRTEffect estimate
Investigator-assessed PFS Full analysis set Full analysis set HR 1.21 (95% CI 0.928–1.573); P = 0.9199

The registry describes the endpoint as time from randomization until documented progressive disease or death, with the analysis extending to a censored date when neither event had occurred and with the registry time frame extending up to 37 months.

Clinical Biostats interpretation

An HR of 1.21 means that the estimated hazard for the avelumab-plus-SOC-CRT group relative to the placebo-plus-SOC-CRT group was 1.21 within the reported time-to-event analysis. Expressed descriptively, the point estimate is above 1 rather than below 1.

This does not mean that 21% more patients progressed or died, nor does it mean that an individual patient's probability of progression or death increased by 21%. A hazard ratio is a relative time-to-event measure, not an absolute risk difference.

The two-sided 95% confidence interval of 0.928 to 1.573 describes the uncertainty around the estimated hazard ratio under the analysis framework. Because the interval spans 1, the estimate is compatible with hazard ratios below 1 as well as above 1.

The reported P = 0.9199 is a measure used for the statistical comparison under the prespecified superiority framework. It is not a measure of effect size, clinical importance, or the probability that the treatment is effective or ineffective.

As with other hazard-ratio analyses, interpretation also depends on the time-to-event framework, censoring, and the extent to which a single relative hazard summarizes the treatment comparison over follow-up. The ClinicalTrials.gov record does not report a proportional-hazards diagnostic or a separate assessment of that assumption.

Educational note: a Kaplan-Meier curve is not reconstructed here from the summary statistics. A valid curve reconstruction requires underlying event and censoring information or sufficiently detailed source data.

7. Secondary Endpoint Results

The registry reports four secondary statistical analyses. Three are time-to-event endpoints analyzed with the log-rank test and summarized by hazard ratios. ORR is a binary endpoint analyzed with the Cochran-Mantel-Haenszel test and summarized by an odds ratio.

Overall Survival

Hazard ratio for death

1.31

95% CI: 0.927–1.849   ·   P = 0.9372

Two-sided confidence interval  ·  Superiority hypothesis

Clinical Biostats interpretation

The OS hazard ratio of 1.31 is the estimated relative hazard for death comparing avelumab plus SOC CRT with placebo plus SOC CRT in the full analysis set. The point estimate is above 1.

The 95% CI of 0.927–1.849 includes 1, so the reported interval is compatible with both a lower and a higher hazard under the treatment-group comparison. The confidence interval therefore communicates substantial uncertainty around the point estimate.

The reported P = 0.9372 is evidence from the specified statistical comparison, not a measure of how large the treatment effect is. It should be interpreted alongside the hazard ratio and confidence interval rather than substituted for them.

Time to Locoregional Failure

Hazard ratio

1.25

95% CI: 0.930–1.694   ·   P = 0.9316

Two-sided confidence interval  ·  Superiority hypothesis

Clinical Biostats interpretation

The hazard ratio of 1.25 is the estimated relative hazard for the time-to-locoregional-failure endpoint in the avelumab-plus-SOC-CRT group compared with the placebo-plus-SOC-CRT group.

The 95% CI of 0.930–1.694 crosses 1. The interval therefore does not identify a single direction of the underlying hazard ratio with the precision implied by an interval entirely on one side of 1.

The reported P = 0.9316 should not be interpreted as a percentage measure of treatment effect. The endpoint is also subject to the usual time-to-event considerations of censoring and the interpretation of a relative hazard over follow-up.

Objective Response Rate

Odds ratio for objective response

0.947

95% CI: 0.663–1.352   ·   P = 0.6229

Two-sided confidence interval  ·  Superiority hypothesis

Clinical Biostats interpretation

An OR of 0.947 compares the odds of objective response between the two randomized groups under the Cochran-Mantel-Haenszel analysis. An odds ratio of 1 would indicate equal odds under the comparison's defined direction.

The 95% CI of 0.663–1.352 includes 1, indicating uncertainty that spans both sides of the null value. The odds ratio should not be read as a probability ratio; odds and probabilities are different quantities.

The reported P = 0.6229 describes the statistical comparison and does not itself quantify the magnitude or clinical importance of the observed odds ratio.

Time to Distant Metastatic Failure

Hazard ratio

1.21

95% CI: 0.909–1.624   ·   P = 0.9061

Two-sided confidence interval  ·  Superiority hypothesis

Clinical Biostats interpretation

The reported HR of 1.21 is the estimated relative hazard for time to distant metastatic failure in the avelumab-plus-SOC-CRT group compared with the placebo-plus-SOC-CRT group.

The 95% CI of 0.909–1.624 includes 1, so the interval encompasses both directions of the relative hazard comparison. The width of the interval also illustrates why the point estimate should not be treated as a precise description of the underlying treatment effect.

The reported P = 0.9061 is a test statistic result, not a measure of effect magnitude. Interpretation should consider the hazard ratio, its confidence interval, the endpoint definition, censoring, and the randomized analysis population together.

8. Results Summary

EndpointMethodEffect measureEstimate95% CIP-value
Primary: PFSLog-rankHazard ratio1.210.928–1.5730.9199
OSLog-rankHazard ratio1.310.927–1.8490.9372
Time to locoregional failureLog-rankHazard ratio1.250.930–1.6940.9316
ORRCochran-Mantel-HaenszelOdds ratio0.9470.663–1.3520.6229
Time to distant metastatic failureLog-rankHazard ratio1.210.909–1.6240.9061

Across the five posted statistical analyses, the ClinicalTrials.gov record shows hazard-ratio point estimates above 1 for all four time-to-event comparisons and an odds-ratio point estimate of 0.947 for objective response. All five reported two-sided confidence intervals include the corresponding null value of 1.

Important distinction: these observations describe the numerical results reported in the ClinicalTrials.gov record. They should not be converted into an overall treatment ranking or a claim about clinical appropriateness. The statistical evidence for each endpoint should be considered according to its prespecified role, analysis method, effect measure, confidence interval, and hypothesis framework.

9. Safety: Serious Adverse Events

The ClinicalTrials.gov record reports serious adverse events by randomized arm using affected participants over participants at risk. No additional safety endpoints are provided in the ClinicalTrials.gov record, so the safety summary is limited to this measure.

Safety measureAvelumab + Standard of Care ChemotherapyPlacebo + SOC CRT
Serious adverse events, affected / at risk184 / 348177 / 344

The denominators are not the same as the total enrollment of 697 because the reported safety measure is expressed using its own affected and at-risk counts. The ClinicalTrials.gov record does not provide a formal statistical comparison for serious adverse events, so no additional inferential conclusion is added.

10. Statistical Methods Explained

Why was a log-rank test used for PFS?

PFS is a time-to-event endpoint. Participants can experience progression or death at different times, while others may remain event-free when their observed follow-up ends. The log-rank test is designed to compare the survival distributions of two groups while incorporating the timing of events and censoring rather than reducing the endpoint to a single binary outcome.

What does the PFS hazard ratio of 1.21 mean?

The HR of 1.21 is the reported relative hazard estimate for PFS comparing avelumab plus SOC CRT with placebo plus SOC CRT. It is not a 21% difference in the proportion of participants who progressed, and it is not an individual-level probability. It summarizes the relative event rate within the time-to-event analysis framework.

Why is the confidence interval important?

A point estimate is only one estimate from the data. The reported 95% CI describes statistical uncertainty around that estimate. For PFS, the interval is 0.928–1.573, which spans 1. This means the plausible range represented by the interval includes both a hazard ratio below 1 and one above 1 under the stated inferential framework.

Why doesn't the P-value measure effect size?

A P-value addresses the statistical evidence against a specified null hypothesis under the analysis model. It does not tell us how large the treatment effect is. The effect size is conveyed by the hazard ratio or odds ratio, while the confidence interval provides information about uncertainty around that effect estimate.

Why was the Cochran-Mantel-Haenszel test used for ORR?

ORR is a binary endpoint rather than a time-to-event endpoint. A Cochran-Mantel-Haenszel approach provides a way to compare categorical outcomes within the trial's statistical framework. The corresponding effect measure reported in the registry is an odds ratio rather than a hazard ratio.

Why does the full analysis set matter?

The registry states that the FAS included all randomized participants. An analysis anchored to randomization preserves the comparison created by random allocation. Restricting efficacy analysis only to participants who remained on treatment could introduce differences related to treatment discontinuation or post-randomization behavior that randomization itself did not balance.

What does censoring mean for these endpoints?

For a time-to-event endpoint, a participant who has not experienced the specified event by the end of observed follow-up does not simply count as an event-free participant forever. Instead, the participant contributes information up to the censoring time. Kaplan-Meier and related survival methods are designed to incorporate that partial follow-up.

11. Understanding the Primary Result More Carefully

Relative effect

The primary PFS estimate was an HR of 1.21. This is a relative time-to-event measure, not an absolute difference in the percentage of participants who experienced progression or death.

Uncertainty

The 95% CI of 0.928–1.573 gives a substantially broader range than the point estimate alone. Reporting the interval prevents the HR from being interpreted as a precise fixed quantity.

Statistical test

The reported P-value was 0.9199 from the log-rank analysis. It describes the statistical comparison under the specified hypothesis and does not measure the size of the observed HR.

Analysis population

The registry states that the FAS included all randomized participants. This is the population used for the reported primary and secondary efficacy analyses.

The most statistically informative way to read the primary result is therefore as a package: the endpoint definition establishes what event was being measured; the Kaplan-Meier/log-rank framework accounts for event timing and censoring; the hazard ratio summarizes the relative treatment comparison; the confidence interval communicates uncertainty; and the P-value describes the formal statistical test.

12. Confidence Intervals Across the Reported Analyses

EndpointEstimateLower limitUpper limitNull value included?
PFS1.210.9281.573Yes
OS1.310.9271.849Yes
Time to locoregional failure1.250.9301.694Yes
ORR0.9470.6631.352Yes
Time to distant metastatic failure1.210.9091.624Yes

For both hazard ratios and odds ratios, the null value is 1. A confidence interval containing 1 indicates that the interval includes equality between the groups on the reported relative-effect scale. This does not mean that the groups are proven identical; it means that the interval does not exclude the null value at the stated confidence level.

13. Effect Measures: Hazard Ratio vs Odds Ratio

FeatureHazard ratioOdds ratio
Used in this trial forPFS, OS, time to locoregional failure, time to distant metastatic failureObjective response rate
Outcome structureTime-to-eventBinary
Null value11
Time information retained?YesNo
Example from this trialPFS HR 1.21ORR OR 0.947

These effect measures should not be placed on the same scale conceptually. A hazard ratio incorporates event timing and censoring through survival analysis, whereas an odds ratio compares the odds of a binary outcome under the specified analysis. The numerical value of one cannot be interpreted identically across all statistical measures even though the null value happens to be 1 for both.

14. Multiplicity and the Superiority Framework

The ClinicalTrials.gov record identifies superiority as the hypothesis type for the primary and secondary statistical analyses. The primary endpoint is PFS, while OS, time to locoregional failure, ORR, and time to distant metastatic failure are secondary endpoints.

Endpoint roleEndpointHypothesis typeMethod
PrimaryPFSSuperiorityLog-rank
SecondaryOSSuperiorityLog-rank
SecondaryTime to locoregional failureSuperiorityLog-rank
SecondaryORRSuperiorityCochran-Mantel-Haenszel
SecondaryTime to distant metastatic failureSuperiorityLog-rank

Multiple endpoints create a broader inferential context than a single hypothesis test. The ClinicalTrials.gov record identifies the endpoint roles and methods but do not provide an alpha-allocation or multiplicity-adjustment strategy. Consequently, this page does not infer a familywise error-control procedure from the reported P-values.

15. Blinding and Randomization as Statistical Design Features

The trial is registered as randomized and quadruple-masked. Randomization is fundamental because treatment assignment is determined independently of participants' eventual outcomes, providing the basis for an unbiased comparison in expectation. Quadruple masking is likewise a design feature intended to reduce knowledge of treatment assignment among relevant trial participants or personnel.

These features do not guarantee identical observed outcomes or eliminate every source of uncertainty. Rather, they establish the framework in which differences between randomized groups can be analyzed without treating post-randomization differences as though they were created by the investigators.

Randomization principle
Treatment assignment → comparable groups in expectation → outcome comparison

The key statistical benefit of randomization is protection against systematic baseline confounding in the treatment comparison. The actual observed groups can still differ by chance, and the ClinicalTrials.gov record does not provide a baseline-characteristics table.

16. What the Registry Does Not Provide in the Supplied Data

The ClinicalTrials.gov record contains the primary and secondary formal statistical analyses but do not provide several types of detail that would be needed for a more extensive reconstruction of the trial's statistical history.

Why this matters: absence of a reported detail in the ClinicalTrials.gov record should not be converted into a claim that the underlying protocol or statistical analysis plan did not contain that detail. This page simply does not infer information that was not reported.

17. Limitations

18. Why This Trial Matters Statistically

JAVELIN Head and Neck 100 is a useful teaching example because its registry results place several core clinical-trial concepts side by side. The primary endpoint is a time-to-event outcome analyzed with Kaplan-Meier and log-rank methods, while a secondary binary endpoint uses the Cochran-Mantel-Haenszel framework and an odds ratio. The same randomized comparison therefore requires different statistical tools depending on the structure of the endpoint.

ConceptHow it appears in this trial
RandomizationThe study is registered as randomized.
BlindingThe study is registered as quadruple-masked.
Kaplan-Meier estimationUsed for the primary PFS analysis.
Log-rank testUsed for PFS, OS, time to locoregional failure, and time to distant metastatic failure.
Hazard ratioUsed as the effect measure for four time-to-event analyses.
Confidence intervalTwo-sided 95% CIs are reported for all five statistical analyses.
Odds ratioUsed for the binary ORR endpoint.
Cochran-Mantel-Haenszel testUsed for ORR.
Full analysis setAll reported efficacy analyses used the FAS, defined as all randomized participants.
Superiority testingThe analyses posted on ClinicalTrials.gov identify superiority as the hypothesis type.

The particularly important lesson is that a clinical-trial result is not just a P-value. A statistically literate reading starts with the estimand-like question embodied in the endpoint, identifies the analysis population, examines the effect measure, reads the confidence interval, and then considers what the hypothesis test contributes to that interpretation.

19. A Structured Reading of the Five Reported Analyses

StepQuestionJAVELIN Head and Neck 100
1What was measured?PFS, OS, time to locoregional failure, ORR, and time to distant metastatic failure.
2Who was analyzed?Full analysis set including all randomized participants.
3What statistical family was used?Survival analysis for four time-to-event endpoints; categorical-data analysis for ORR.
4What was the effect measure?Hazard ratio for time-to-event endpoints; odds ratio for ORR.
5What is the estimate?HR 1.21 for PFS; HR 1.31 for OS; HR 1.25 for time to locoregional failure; OR 0.947 for ORR; HR 1.21 for time to distant metastatic failure.
6How precise is it?Each reported 95% CI includes 1.
7What is the formal test result?P-values range from 0.6229 to 0.9372 across the five reported analyses.

This framework avoids a common mistake in trial interpretation: starting with the P-value and working backward. The endpoint and analysis method come first. The effect estimate and confidence interval then describe the observed treatment comparison and its uncertainty, while the P-value supplies the formal test result.

20. Clinical Interpretation vs Statistical Interpretation

Statistical interpretation

The registry analyses report relative effect estimates for five efficacy endpoints. The primary PFS HR is 1.21 with a two-sided 95% CI of 0.928–1.573 and P = 0.9199. The four secondary analyses likewise report confidence intervals that include the null value of 1.

Clinical interpretation

The registry data alone do not provide median survival, absolute survival estimates, response percentages, subgroup results, or other clinical-effect summaries needed for a broader clinical interpretation. Those quantities are therefore not inferred here.

This distinction is important. Statistical analysis can precisely describe what the registry-reported estimates say without claiming more than the data support. In particular, a hazard ratio should not be converted into an absolute benefit, and the absence of an interval excluding 1 should not be transformed into a claim that two treatments are identical.

21. Statistical Concepts in This Trial

Learn more about the methods used in this trial:

22. Related Statistical Calculators

Use these related tools to explore the statistical quantities that appear in the reported analyses:

23. Sources

Continue through Clinical Biostats

Explore the statistical concepts behind randomized clinical trials, time-to-event endpoints, categorical outcomes, and treatment-effect measures.

24. Record Summary

JAVELIN Head and Neck 100 provides a compact example of how different clinical-trial endpoints require different statistical approaches. The primary endpoint, investigator-assessed PFS, is a time-to-event outcome analyzed with Kaplan-Meier and log-rank methods and summarized with a hazard ratio. OS, time to locoregional failure, and time to distant metastatic failure use the same broad survival-analysis framework. ORR, by contrast, is a binary endpoint analyzed using the Cochran-Mantel-Haenszel test and summarized by an odds ratio.

The ClinicalTrials.gov record reports a PFS HR of 1.21 (95% CI 0.928–1.573; P = 0.9199), an OS HR of 1.31 (95% CI 0.927–1.849; P = 0.9372), a time-to-locoregional-failure HR of 1.25 (95% CI 0.930–1.694; P = 0.9316), an ORR OR of 0.947 (95% CI 0.663–1.352; P = 0.6229), and a time-to-distant-metastatic-failure HR of 1.21 (95% CI 0.909–1.624; P = 0.9061).

The most useful statistical reading is not any single number in isolation. It is the combination of endpoint definition, randomized analysis population, appropriate statistical method, effect estimate, confidence interval, and hypothesis-test result. The ClinicalTrials.gov record supports that structured interpretation without requiring unsupported assumptions about median survival, subgroup effects, multiplicity, or other unreported design details.

Clinical Biostats methodology: A trial-results page should distinguish reported evidence from statistical explanation. The objective is to make the statistical structure of the trial understandable while preserving the exact reported estimates and avoiding conclusions that depend on information not contained in the ClinicalTrials.gov record.